The past few years has seen great advancement in the world of generative models (GANs), becoming one of the most promising approaches toward collecting all of the easily accessible open-source information available and using it to develop models and algorithms to analyze and understand.  

The below blog brings together a general explanation of GANs, alongside the newest papers, video presentations, application methods and more.


What are generative models?

For a model to be "Generative" must fit in a class of statistical models which contrast against discriminative models. One of the more informal definitions for this was given by the developer group at Google:

  • Generative models can generate new data instances.
  • Discriminative models discriminate between different kinds of data instances.

Simply, a generative model is one that can generate new data after learning from the dataset. VAE’s & GANs are the most popular types of generative models. Therefore, "Generative" describes a class of statistical models that contrasts with discriminative models.

Informally:

  • Generative models can generate new data instances.
  • Discriminative models discriminate between different kinds of data instances.

A generative model could generate new photos of animals that look like real animals, while a discriminative model could tell a dog from a cat. GANs are just one kind of generative model.

More formally, given a set of data instances X and a set of labels Y:

  • Generative models capture the joint probability p(X, Y), or just p(X) if there are no labels.
  • Discriminative models capture the conditional probability p(Y | X).

Top GANs Video Presentations

Free to view:

Building Generative Models Of Symptomatic Health Data for Autonomous Deep Space Missions - Krittika D'Silva , AI Researcher at NASA Frontier Development Lab

Learning a Phenotype Representation for AI-assisted Leukemia Diagnosis Using Deep Learning Generative Model - Jeremy Lee , Assistant Professor at National Tsing Hua University

A Deep Generative Model Approach To The Genetic Analysis Of Medical Images - Francesco Paolo Casale, Data Scientist at Insitro


Register Free Account to View (no cost):

Using GANs to Estimate Value-at-Risk for Market Risk Management - Hamaad Shah , Vice President - Lead Data Scientist at Deutsche Bank

A Deep Generative Model Approach To The Genetic Analysis Of Medical Images - Francesco Paolo Casale, Data Scientist at Insitro

Learning a Phenotype Representation for AI-assisted Leukemia Diagnosis Using Deep Learning Generative Model - Jeremy Lee , Assistant Professor at National Tsing Hua University

Do Deep Generative Models Know What They Don't Know - Balaji Lakshminarayanan, Staff Research Scientist at DeepMind

The Power of Large scale RL and Generative Models - Ilya Sutskever, Co-Founder & Research Director at OpenAI

New Optimization Perspective On Generative Adversarial Networks - Simon Lacoste-Julien - Simon Lacoste-Julien, VP Lab Director/Associate Professor at SAIT AI Lab/Université de Montréal

Visualizing and Understanding Generative Adversarial Networks - David Bau, PhD Student at MIT CSAIL

Generative Adversarial Networks - Ian Goodfellow, Research Scientist at Google Brain

Building Generative Models Of Symptomatic Health Data for Autonomous Deep Space Missions - Krittika D'Silva , AI Researcher at NASA Frontier Development Lab

Personalized Generative Models - Yaniv Taigman, Research Scientist at Facebook AI Research (FAIR)

Improving Image Classification with Generative Adversarial Network - Michael Dietz, Founder at Waya.ai

Prototype-Based Drug Discovery using Deep Generative Models - Shahar Harel, Research Scientist at Technion - Israel Institute of Technology

Synthesis of Images by Two-Stage Generative Adversarial Networks - Qiang Huang, Senior Researcher at University of Surrey

Application of Generative Adversarial Networks (GANs) in Algorithmic Trading and Aggregation of Low-Alpha Strategies - Mohammad Yousuf Hussain , Senior Technology and Innovation Specialist at HSBC

Panel Discussion: How Can We Overcome Challenges To Fully Leverage Opportunities Of GANs - Prashant Raina, Concordia University; Simon Lacoste-Julien, SAIT AI Lab/Université de Montréal; Bharatendra Rai, University of Massachusetts; Alexia Jolicoeur- Matineau, MILA


New Papers/Work in the Space

The Best Generative Models Papers from the ICLR

Handling incomplete heterogeneous data using VAEs

A deep generative model trifecta: Three advances that work towards harnessing large-scale power

18 Impressive Applications of Generative Adversarial Networks (GANs)

5 applications of generative adversarial networks

Applications of GANs

Generative Adversarial Nets

A Style-Based Generator Architecture for Generative Adversarial Networks

The Secret Revealer: Generative Model-Inversion Attacks Against Deep Neural
Networks


Interest in hearing more on GANs? Join our virtual event!

The Generative Models Stage of RE•WORK's Deep Learning 2.0 Virtual Summit in January is set to cover the most recent industry applications with 50 speakers on the day. You can read more on this here.

https://www.re-work.co/events/generative-models-summit

Further Reading: